SHIPS - Ground Amplification Studies

Barberopoulou, A., A. Qamar, T.L. Pratt, K.C. Creager, and W.P. Steele, 2004, Local amplification of seismic waves from the Mw7.9 Alaska earthquake and damaging seiches in Lake Union, Seattle, Washington, Geophys. Res. Lett., v. 31, L03607, doi: 10.1029/2003GL018569. PDF

The Mw7.9 Denali, Alaska earthquake of 3 November, 2002, caused minor damage to at least 20 houseboats in Seattle, Washington by initiating water waves in Lake Union. These water waves were likely initiated during the large amplitude seismic surface waves from this earthquake. Maps of spectral amplification recorded during the Denali earthquake on the Pacific Northwest Seismic Network (PNSN) strong-motion instruments show substantially increased shear and surface wave amplitudes coincident with the Seattle sedimentary basin. Because Lake Union is situated on the Seattle basin, the size of the water waves may have been increased by local amplification of the seismic waves by the basin. Complete hazard assessments require understanding the causes of these water waves during future earthquakes.

Brocher, T.M., T. L. Pratt, C.S. Weaver, C.M. Snelson, and A.D. Frankel, 2002, Implosion, earthquake, and explosion recordings from the 2000 Seattle Kingdome Seismic Hazards Investigation in Puget Sound (SHIPS), Washington, U.S. Geological Survey Open-File Report 02-123, 29 p. http://geopubs.wr.usgs.gov/open-file/of02-123/

This report describes seismic data obtained in Seattle, Washington, March 24-28, 2000, during a Seismic Hazards Investigation of Puget Sound (SHIPS). The seismic recordings obtained by this SHIPS experiment, nicknamed Kingdome SHIPS, were designed to (1) measure site responses throughout Seattle and to (2) help define the location of the Seattle fault. During Kingdome SHIPS, we recorded the Kingdome implosion, four 150-lb (68-kg) shots, and a M w = 7.6 teleseism using a dense network of seismographs deployed throughout Seattle. The seismographs were deployed at a nominal spacing of 1 km in a hexagonal grid extending from Green Lake in the north to Boeing Field in the south.

The Seattle Kingdome was a domed sports stadium located in downtown Seattle near the Seattle fault. The Seattle Kingdome was imploded (demolished) at 8:32 AM local time (16:32 UTC) on March 26 (JD 086), 2000. The seismic energy produced by implosion of the Kingdome was equivalent to a local earthquake magnitude of 2.3. Strong impacts produced by the implosion of the Kingdome generated seismic arrivals to frequencies as low as 0.1 Hz. An mpeg movie of the ground motions recorded during the demolition of the Kingdome may be downloaded from the following website: ../../../../research/groundmotion/movies/. This movie documents longer shaking durations in the Duwamish River valley, as expected for the low shear wave velocities found in these youthful alluvial deposits along the river.

Although the shots varied in their quality, useful seismic refraction data were acquired from all four shot points, located in the corners of our temporary array. Two shots located north of the Seattle fault, where the charges were detonated within the ground water column (Discovery and Magnuson Parks), were much more strongly coupled than were the two shots to the south of the Seattle fault, where the shots were detonated above the water table (Lincoln and Seward Parks).

Thirty-eight RefTek stations, scattered throughout Seattle, recorded the M w = 7.6 Japan Volcano Islands earthquake (22.4˚N, 143.6˚E, 104 km depth) of 28 March 2000 (JD 088). This teleseism produced useful signals for periods between 4 and 7 seconds. Only a few recordings of small magnitude local earthquakes were made, and these recordings are not presented.

In this report, we describe the acquisition of these data, discuss the processing and merging of the data into common shot gathers, and illustrate the acquired data. We also describe the format and content of the archival tapes containing the SEGY-formatted, common-shot gathers.

Brocher, T.M., Pratt, T.L., and the Dry SHIPS Working Group, 2001, Dry SHIPS Recordings of the Chi-Chi earthquake, Seattle, Washington, Bull. Seismol. Soc. Amer., v. 91, p. 1395 (with CD-supplement of seismic data). PDF

Dry SHIPS was a seismic refraction study of the Seattle basin, Washington State, involving 29 chemical explosions recorded by 1008 seismographs, conducted in September 1999. The E-W trending Dry SHIPS line ran eastward from the Olympic Peninsula, through the Seattle basin, to the foothills of the Cascade mountains. Twenty-eight REFTEKs (DAS Model 07G; 24-bit) deployed along the Dry SHIPS line at 4 km intervals were programmed to record data continuously from 3-component sensors (Table 1). These 28 stations recorded the main shock and several aftershocks of the Mw = 7.6 Chi-Chi earthquake (Shin et al., 2000; Brocher et al., 2000a).

The recordings were made with REFTEKs described by PASSCAL (1991). The three-component Mark Products (copyright) L-28 4.5-Hz geophones were oriented such that the longitudinal (N-S) component was directed to magnetic north. [The eastward declination of magnetic north relative to true north in Seattle is about 20 degrees.] The REFTEKs were equipped with Global Positioning System (GPS) receivers to synchronize the internal timing on the individual REFTEKs to satellite timing. The REFTEKs recorded continuously at a sample rate of 250 samples/sec (4 msec sample interval) but have been resampled here to a 20 msec sample rate after the application of an anti-alias filter. The original records are available from the IRIS Data Management Center in Seattle. On this CD the stations are identified by their Data Acquisition System (DAS) number (Table 1).

The azimuth of propagation to the Dry SHIPS array was about 37 degrees for the Chi-Chi signals. The stations were located at distances between 88.5 and 89.5 degrees from the Chi Chi epicenter. The Dry SHIPS recordings of the Chi-Chi mainshock provide useful signals from 50 Hz down to 10 second periods, although at higher frequencies the Chi-Chi signals are not discernible due to the ambient noise. A further caveat is that these data contain abrupt offsets in amplitude caused by voltage steps associated with writing data to disk. At high frequencies these VOLTAGE steps are lost in the noise and pose no problems, but at longer periods (e.g., greater than 4 seconds) these steps may cause serious problems on all three channels.

The recordings document a significant (factor between 5 and 10) amplification of compressional- and shear-wave energy in the Seattle basin at periods between 1 and 2 seconds relative to BEDROCK sites east and west of the basin (Brocher et al., 2000b). Signal durations in the Seattle basin were also substantially increased relative to BEDROCK sites in the Olympic peninsula and Cascade foothills, approaching 100 seconds in the vicinity of Seattle.

Hartzell, S., Carver, D., Cranswick, E., and A. Frankel, 2000, Variability of site response in Seattle, Washington, Bull. Seism. Soc. Am., 90, 1237-1250. PDF

Ground motion from local earthquakes and the SHIPS (Seismic Hazards Investigation in Puget Sound) experiment is used to estimate site amplification factors in Seattle. Earthquake and SHIPS records are analyzed by two methods: (1) spectral ratios relative to a nearby site on Tertiary sandstone, and (2) a source/site spectral inversion technique. Our results show site amplifications between 3 and 4 below 5 Hz for West Seattle relative to Tertiary rock. These values are approximately 30% lower than amplification in the Duwamish Valley on artificial fill, but significantly higher than the calculated range of 2 to 2.5 below 5 Hz for the till-covered hills east of downtown Seattle. Although spectral amplitudes are only 30% higher in the Duwamish Valley compared to West Seattle, the duration of long-period ground motion is significantly greater on the artificial fill sites. Using a three-dimensional displacement response spectrum measure that includes the effects of ground-motion duration, values in the Duwamish Valley are 2 to 3 times greater than West Seattle. These calculations and estimates of site response as a function of receiver azimuth point out the importance of trapped surface-wave energy within the shallow, low-velocity, sedimentary layers of the Duwamish Valley. One-dimensional velocity models yield spectral amplification factors close to the observations for till sites east of downtown Seattle and the Duwamish Valley, but underpredict amplifications by a factor of 2 in West Seattle. A two-dimensional finite-difference model does equally well for the till sites and the Duwamish Valley and also yields duration estimates consistent with the observations for the Duwamish Valley. The two-dimensional model, however, still underpredicts amplification in West Seattle by up to a factor of 2. This discrepancy is attributed to 3D effects, including basin-edge–induced surface waves and basin-geometry–focusing effects, caused by the proximity of the Seattle thrust fault and the sediment-filled Seattle basin.

Li, Q., W. Wilcock, T.L. Pratt, C.M. Snelson, and T.M. Brocher, 2006, Seismic attenuation structure of the Seattle basin, Washington state, from explosive-source refraction data; Bull. Seism. Soc. Am., 96, in press. PDF

We used waveform data from the 1999 SHIPS (Seismic Hazard Investigation of Puget Sound) seismic refraction experiment to constrain the attenuation structure of the Seattle basin, Washington State. We inverted the spectral amplitudes of compressional- and shear-wave arrivals for source spectra, site responses, and one- and two-dimensional Q -1 models at frequencies between 1 and 40 Hz for P-waves and 1 and 10 Hz for S-waves. We also obtained Q -1 models from t* values calculated from the spectral slopes of P-waves between 10 and 40 Hz. One-dimensional inversions show that Q p at the surface is 22 at 1 Hz, 130 at 5 Hz and 390 at 20 Hz. The corresponding values at 18 km depth are 100, 440 and 1900. Q s at the surface is 16 and 160 at 1 Hz and 8 Hz, respectively, increasing to 80 and 500 at 18 km depth. The t* inversion yields a Q P model that is consistent with the amplitude inversions at 20 and 30 Hz. The basin geometry is clearly resolved in the t* inversion, but the amplitude inversions only imaged the basin structure after removing anomalously high amplitude shots near Seattle. When these shots are removed, we infer that Q -1 values may be ~30% higher in the center of the basin than the one-dimensional models predict. We infer that seismic attenuation in the Seattle basin will significantly reduce ground motions at frequencies at and above 1 Hz, partially countering amplification effects within the basin.

Pitarka, A., R. Graves, and P. Somerville, 2004, Validation of a 3D velocity model of the Puget Sound region based on modeling ground motion from the February 28, 2001 Nisqually earthquake, Bull. Seism. Soc. Am., 94, 1670-1689. PDF

In this study we prepared a 3D velocity model suitable for modeling long-period wave propagation in the Puget Sound region. The model is based on products of the Seismic Hazard Investigation in Puget Sound (SHIPS) and geophysical information from other studies of the region. The adequacy of the velocity model was evaluated based on analyses of goodness of fit between recorded and simulated ground-motion velocity from the M 6.8 Nisqually earthquake. The earthquake was located about 60 km south of Seattle with a hypocentral depth of 59 km. The analyses were performed in the frequency range of 0.02–0.5 Hz, using data from 40 stations. Although our model covers a wide area of the Puget Sound region, its quality is assessed in the Seattle region in which the distribution of stations that recorded the Nisqually earthquake was denser. Our 3D finite-difference ground-motion modeling suggests that the propagation of long-period waves (periods longer than 3 sec) in the Seattle basin is mostly affected by the deep basin structure. The tomographic velocity model of Parsons et al. (2001), combined with the model of depth to the basement of the Seattle basin of Blakelyet al. (1999), was essential in preparing and constraining geometrical features of the proposed velocity model.

Pratt, T.L., K.L. Meagher, T.M. Brocher, T. Yelin, R. Norris, L. Hultgrien, E. Barnett, and C.S. Weaver, 2003, Earthquake recordings from the 2002 Seattle Seismic Hazard Investigation of Puget Sound (SHIPS), Washington State, U.S. Geological Survey Open-File Report 03-361, 72 p. http://geopubs.wr.usgs.gov/open-file/of03-361/

This report describes seismic data obtained during the fourth Seismic Hazard Investigation of Puget Sound (SHIPS) experiment, termed Seattle SHIPS. The experiment was designed to study the influence of the Seattle sedimentary basin on ground shaking during earthquakes. To accomplish this, we deployed seismometers over the basin to record local earthquakes, quarry blasts, and teleseisms during the period of January 26 to May 27, 2002. We plan to analyze the recordings to compute spectral amplitudes at each site, to determine the variability of ground motions over the basin. During the Seattle SHIPS experiment, seismometers were deployed at 87 sites in a 110-km-long east-west line, three north-south lines, and a grid throughout the Seattle urban area (Figure 1). At each of these sites, an L-22, 2-Hz velocity transducer was installed and connected to a REF TEK Digital Acquisition System (DAS), both provided by the Program for Array Seismic Studies of the Continental Lithosphere (PASSCAL) of the Incorporated Research Institutes for Seismology (IRIS). The instruments were installed on January 26 and 27, and were retrieved gradually between April 18 and May 27. All instruments continuously sampled all three components of motion (velocity) at a sample rate of 50 samples/sec. To ensure accurate computations of amplitude, we calibrated the geophones in situ to obtain the instrument responses. In this report, we discuss the acquisition of these data, we describe the processing and merging of these data into 1-hour long traces and into windowed events, we discuss the geophone calibration process and its results, and we display some of the earthquake recordings.

Pratt, T.L., T.M. Brocher, C.S. Weaver, K.C. Creager, C.M. Snelson, R.S. Crosson, K.C. Miller, and A.M. Tréhu, 2003, Amplification of seismic waves by the Seattle basin, Bull. Seism. Soc. Am., 93, 533-545. PDF

Recordings of the 1999 Mw 7.6 Chi-Chi ( Taiwan) earthquake, two local earthquakes, and five blasts show seismic-wave amplification over a large sedimentary basin in the U.S. Pacific Northwest. For weak ground motions from the Chi-Chi earthquake, the Seattle basin amplified 0.2- to 0.8-Hz waves by factors of 8 to 16 relative to bedrock sites west of the basin. The amplification and peak frequency change during the Chi-Chi coda: the initial S-wave arrivals (0–30 sec) had maximum amplifications of 12 at 0.5–0.8 Hz, whereas later arrivals (35–65 sec) reached amplifications of 16 at 0.3–0.5 Hz. Analysis of local events in the 1.0- to 10.0-Hz frequency range show fourfold amplifications for 1.0-Hz weak ground motion over the Seattle basin. Amplifications decrease as frequencies increase above 1.0 Hz, with frequencies above 7 Hz showing lower amplitudes over the basin than at bedrock sites. Modeling shows that resonance in low-impedance deposits forming the upper 550 m of the basin beneath our profile could cause most of the observed amplification, and the larger amplification at later arrival times suggests surface waves also play a substantial role. These results emphasize the importance of shallow deposits in determining ground motions over large basins.

Pratt, T.L., and T.M. Brocher, 2005, Attenuation within sedimentary basins and the shapes of site response curves in the Puget Lowland, Washington State, Bull. Seism. Soc. Am., in press.

Simple spectral ratio (SSR) and horizontal-to-vertical (H/V) site response estimates at 47 sites in the Puget Lowland of Washington State document significant attenuation of 1.5 to 20 Hz shear waves within sedimentary basins there. Amplitudes of the horizontal components of shear-wave arrivals from three local earthquakes were used to compute SSRs with respect to the average of two bedrock sites, and H/V spectral ratios with respect to the vertical component of the shear-wave arrivals at each site. SSR site response curves at thick basin sites show peak amplifications of 2 to 6 at frequencies of 3 to 6 Hz, and decreasing spectral amplification with increasing frequency above 6 Hz. SSRs at non-basin sites show a variety of shapes and larger resonance peaks. We attribute the spectral decay at frequencies above the amplification peak at basin sites to attenuation within the basin strata. Computing the frequency-independent, depth-dependent attenuation factor (Q s,int) from the SSR spectral decay between 2 to 20 Hz gives values of 5 to 40 for shallow sedimentary deposits and about 250 (error range of 90-1000) for the deepest sedimentary strata. H/V site responses show less spectral decay than the SSR responses but contain many of the same resonance peaks. We hypothesize that the H/V method yields a flatter response than SSRs across the frequency spectrum because the H/V reference signal (vertical component of the shear-wave arrivals) has undergone a similar degree of attenuation as the horizontal component recordings. Correcting the SSR site responses for attenuation within the basins improvesagreement between SSR and H/V estimates.

Pratt, T.L., M.E. Templeton, R. Frost, and A. Pierson, 2006, Variations in geophone responses in temporary seismic arrays, Seism. Res. Lett., 76, in press. PDF

Geophones with resonant frequencies of 1 to 4 Hz are commonly used short-period sensors for site response studies and crustal reflection, refraction and tomography experiments. In many cases these studies require only accurate arrival time picks and not amplitude information, but site response and attenuation studies require accurate amplitude (ground velocity) data from these sensors. Furthermore, geophones are often used in these studies to record seismic phases at or below their nominal resonant frequency.
For studies that require accurate ground velocity measurements or use a variety of sensor types, the instrument response must be removed from the data using either the manufacturer’s specifications or results from geophone calibrations. Geophone placement, particularly any tilt to the sensor, can alter the instrument response significantly. The geophone response ideally is calibrated while the sensors are in the ground, but such in situ calibrations generally are not carried out because of time constraints. More commonly, it is simply assumed the geophones have the response characteristics specified by the manufacturer.
In a recent site response study in the Seattle area, named the Seattle Seismic Hazard Investigation of Puget Sound, we used 72 three-component geophones from the Program for Array Seismic Studies of the Continental Lithosphere (PASSCAL) instrument pool to determine the amplitudes of both horizontal and vertical ground shaking at sites over the Seattle sedimentary basin. During the experiment, we used the signal-coil calibration method to measure, in situ, the geophone response of all three components of each sensor (216 channels total). Also during these calibrations, ten geophones were calibrated twice without being disturbed. We subsequently calibrated all of the geophones immediately upon their return to the PASSCAL instrument center using the same calibration method.
Assuming that calibration tests give an accurate measure of the geophone response, the calibrations show the variation in geophone response that can be expected when installing sensors in a typical temporary deployment. Furthermore, the results answer the question of whether calibrations done in a laboratory can be used to correct the instrument response when the geophones are deployed. We conclude that the vertical component is relatively insensitive to the geophone placement, but in situ calibrations are mandatory if horizontal ground motions are to be known to within 10%. This is especially true for signals below the fundamental frequency, where the geophones showed large variations in response during a typical deployment.

Back to Top